Dynamic Contingency Analysis Tool
(DCAT) – A Framework for Analysis of
Extreme Events in Power Systems
January 13, 2017 1
LANL 2017 Grid Science Winter School & Conference
Santa Fe, NM
January 8-13, 2017
Mallikarjuna Vallem, PhD
Senior Research Engineer
Electricity Infrastructure/Analytics
Pacific Northwest National Laboratory
Richland, WA
PNNL Team Nader Samaan (Principal Investigator) Jeff Dagle (Project Manger) Yuri Makarov (Chief Scientist) Ruisheng Diao Tony Nguyen Laurie Miller Bharat Vyakaranam Shaobu Wang Frank Tuffner Prof. MA Pai (Professor Emeritus, Electrical and Computer Engineering, University of Illinois at Urbana-Champaign, Illinois)
Challenges Facing the Power Industry in
Performing Cascading Outage Analysis
2
Results of a recent IEEE Cascading Failure Working Group
survey on the Analysis of Cascading Outages show the
challenges facing the power industry:
70% of the responders indicated that cascading outage analysis in
their organization is not an automated process.
50% of the responders were not satisfied with currently available
tools for analysis of cascading outages.
Dynamic simulation of cascading outages, which should include
protection system modeling, was cited as the most critical feature
that present tools fail to address.
January 13, 2017 |
How Can DCAT Bridge These Gaps?
3
DCAT attempts to bridge multiple gaps in cascading-outage
analysis in a single, unique prototype tool capable of
automatically simulating and analyzing cascading sequences in
real systems using multiprocessor computers.
It uses a hybrid dynamic and steady-state approach to simulating the
cascading outage sequences that includes both fast dynamic and slower
steady-state events.
It integrates dynamic models with protection scheme models for
generation, transmission, and load.
It models special protection systems (SPS)/remedial action schemes
(RAS) and automatic and manual corrective actions.
The current prototype DCAT implementation has been
developed as a Python code that accesses the simulation
functions of the Siemens PSS®E planning tool (PSS/E).
January 13, 2017 |
Overall DCAT Project Phase I Framework
4
Fram
ewo
rk
Survey existing approaches, industry practice, tools, and gaps in performing cascading outage analysis. (Sec. 1)
Develop a hybrid dynamic and steady-state approach to mimic the cascading failure process that includes both fast dynamic and slower events. (Sec. 2)
Integrate dynamic models with protection scheme models. Use NERC standards for relay settings. (Sec. 3)
Perform post-dynamic analysis that models SPS/RAS, automatic and manual corrective actions. (Sec. 4)
Run simulations for the test and full interconnection system models to demonstrate key concepts of the DCAT project. (Sec. 5)
Perform steady-state cascading outage analysis using TransCARE for filtering initiating events and implementing and testing the concept of critical events corridors. (Sec. 6)
Share lessons learned and suggest a roadmap for further improvements. Provide technology outreach and dissemination. (Sec. 7)
January 13, 2017 |
Industry Partnership
Siemens PTI
PSS/E is used for dynamic/steady-state analysis of cascading events.
Multiple copies of PSS/E were provided to develop high performance
computing capability for cascading events.
Provided help with protection modeling.
Provided consulting support for using advanced features in PSS/E.
EPRI
TransCARE is used for steady-state-based cascading outage analysis
for prescreening of initiating events.
ERCOT
Played advisory role through biweekly web conferences to discuss
issues and progress.
DCAT creates an open architecture for extreme events analysis. Any
software vendor meeting technical requirements can connect.
5 January 13, 2017 | 5
DCAT Computational Flow Chart (part A)
6
DCAT Computational Flow Chart (part B)
7 7
Protection Modeling in Phase I
Need to model protection action in order to simulate
cascading events
Used a subset of the protection device models in PSS/E
Generator bus protection (NERC Standard PRC-024-1)
Under-voltage and over-voltage
Under-frequency and over-frequency
Out of step (user-defined model that excludes
nonsynchronous machines)
Load shedding
Under-frequency (frequency responsive non-firm load)
Under-frequency and under-voltage firm load shedding
Transmission protection
Distance relay protection (dynamic simulation)
Overcurrent protection (steady-state simulation)
Breaker location for placement of protection within the
transmission network and the associated operation of zones of
protection
Modeling of RAS/SPS in the steady-state post-dynamic
analysis
8 January 13, 2017 |
Verification if System Reaches a Stable
Point at the end of Dynamic Simulation
Dynamic simulation is a
computationally intensive task. An
appropriate trade-off is necessary
to run the dynamic simulation long
enough to capture the dynamic
response of the system.
The appropriate time can be
determined by having stability
checks at intermediary times that
could stop the dynamic simulation.
The simulation can initially run for
30 seconds. After 30 seconds,
every 5 seconds, values of
standard deviation of the speeds of
all generators are compared with a
certain tolerance. 9 January 13, 2017 |
DCAT Post-Dynamic Simulation
Analysis – Extracting a Steady-State
Case
10
Pre-Dynamic
Post-Dynamic
Newton-Raphson
Power Flow
Post-Dynamic INLF
(Inertial Response)
Bus # Bus Name Bus Voltage Bus Voltage Absolute
Difference Bus Voltage Absolute Difference
Mag (pu) Angle
(deg)
Mag
(pu)
Angle
(deg) Δmag Δangle
Mag
(pu)
Angle
(deg) Δmag Δangle
3018 CATDOG_G 1.0218 -4.08 1.0218 -4.08 0 0 1.0217 -4.08 1E-04 0
January 13, 2017 |
DCAT Post-Dynamic Simulation
Analysis – Corrective Actions
Extreme events could result in a
significant power mismatch between
generation and load. Consequently, a
power flow solution might not converge
without appropriate setup.
As part of DCAT methodology, after
performing a dynamic simulation,
corrective actions could be taken by
the operator to make sure that voltage
and flow violations are reduced.
These actions are
SPS/RAS
automatic actions for voltage violation
corrections
generation redispatch and load
shedding to avoid the tripping of
overloaded lines.
11
Dyn
amic
Sim
ula
tio
ns
Emergency control of power
systems
12
Power systems operates with defined
system conditions
Line ratings
Voltage limits
Generator real and reactive power dispatch
Generator ramp limits
System is considered to operate in
“Emergency condition” if some of the
operational limits are violated
Corrective control actions can be
performed to bring the system out of
“Emergency condition”
Control actions
Preventive
Corrective
Fig.1: Power systems operational states
13
Emergency control actions can be formulated as optimization problem
Minimize corrective actions to mitigate emergency conditions
Where Ci is the weight coefficient, and ∆𝐴𝑟 set of available actions:
2
1...
minr r
r Na
C A
Emergency control of power systems:
Mathematical formulation
MW Dispatch – fixed gen voltage
setpoints
MVar Dispatch – Voltage setpoints
change
Capacitor and Reactor Switching
Transformer Tap Change
Phase Shifter settings
Line Switching (In and Out), including
Switching Not Affected Lines
Load Curtailment Fig. 2- A conceptual view of the Coordinated Emergency Control and Protection System
14
Emergency control actions can be formulated as optimization problem
Minimize corrective actions to mitigate emergency conditions
Subjected to:
Power balance equations
Line flow limits
Bus voltage limits
Generator real and reactive power limits
Additional constraints:
Constraint that describes maximum allowable number of corrective actions
Constraint for linearized operational boundaries
2
1...
minr r
r Na
C A
m axN a N a
1
, for all i
Nb
ij j ij j i
j
P Q
Emergency control of power systems:
Mathematical formulation
Example 1 – Line Fault with a Pilot Scheme
(Transfer Trip Enabled)
January 13, 2017 15
A line fault is applied on one of the lines connected to Bus X1 at a distance
of 90% from the bus. Distance relays are enabled on both ends of the line
with an ability to send a transfer trip to the other end upon sensing a Zone 1
fault.
Though the other end of the line (at Bus X1) sees a Zone 2 fault, this pilot
scheme trips the breaker as soon as the other relay times out on the Zone 1
fault.
Upon successful operation of both breakers, the fault is isolated and no
other tripping actions occur.
Bus X1 Bus Y1
Bu
ses
X1
an
d Y
1
Vo
ltag
es in
pu
Example 2 – Line Fault with No Pilot Scheme
January 13, 2017 16
Same as Example 1, but the communication channel for transfer trip failed.
As a result of that, the near end of the line to the fault at Bus Y1 trips on the
Zone 1 setting (4 cycles) and the other end of the line trips at the Zone 2
setting (22 cycles).
Since the Zone 2 trip persists longer than the Zone 1 trip, many other
relays would have their timers started and some would have cascaded
trips.
Channel Plot
Time (seconds)
5.55.2554.75
1.25
1
0.75
0.5
0.25
0
Bu
ses
X1
an
d Y
1
Vo
ltag
es in
pu
Nea
rby
Gen
erat
or
Bu
s V
olt
age
Gen. under voltage tripping
Example 2 – Line Fault with No Pilot Scheme (2)
January 13, 2017 17
Channel Plot
Time (seconds)
6.756.56.2565.755.55.2554.75
1.25
1
0.75
0.5
0.25
0
Generator Over-Voltage Trippings for Example 2
Generator Under-Frequency Tripping for Example 2
Example 2 – Results Summary
January 13, 2017 18
Generation Loss (MW) 3,004
Load Loss (MW) 0
# of Total Tripping Actions 18
# of Special Protection Systems Triggered 0
# of Overloaded Lines 0
Corrective Actions None
DISTR1 TimeOut Busfrom Busto Ckt
DISTR1 5.05 X1 X2 1
DISTR1 5.333 Y1 Y2 1
VTGTPA TimeOut Pgen Qgen
VTGTPA 5.387 1205 157
VTGTPA 5.387 1195 153
VTGTPA 6.421 68 28.8
VTGTPA 6.421 67 28.8
VTGTPA 6.487 17 14.8
VTGTPA 6.487 17 14.8
VTGTPA 6.571 15 0
VTGTPA 6.579 69 6.59
VTGTPA 6.583 71 6.59
VTGTPA 6.583 70 6.78
VTGTPA 6.583 68 6.59
FRQTPA TimeOut Pgen Qgen
FRQTPA 9.662 7.53 7.37
FRQTPA 9.662 5.42 0
OutOfStep_new TimeOut Pgen Qgen AngleThr AngleDev
OutOfStep_new 10.1374 0 -12.25 180 180.2261
VTGTPA 16.046 74.67 38
VTGTPA 16.046 71.61 38
Example 3 – Demonstration of Under-Frequency
Load Shedding after an Extreme Event
January 13, 2017 19
A bus fault that lasted for six-cycles was introduced at a large substation.
All elements connected to this substation were then tripped to isolate the fault.
This was one of the extreme events that had the potential for many demand-
response–based under-frequency load relays to act and shed a part of the load
dynamically.
A significant amount of generation was lost due to this fault, which was followed by
many under-frequency load sheddings.
At the end, the system was able to reach a stable point.
Generation Loss (MW) 3,900
Load Loss (MW) 1,067
# of Total Tripping Actions 84
# of Special Protection Systems Triggered 0
# of Overloaded Lines 0
Corrective Actions None
Example 3 – Demonstration of Under-Frequency
Load Shedding after an Extreme Event
January 13, 2017 20
DISTR1 TimeOut From To
DISTR1 10.054 X1 Y1
DISTR1 10.054 X2 Y2
FRQTPA TimeOut Pgen Qgen
FRQTPA 10.104 70.56 -10.05
FRQTPA 10.104 70.56 -10.05
OutOfStep_new TimeOut Pgen Qgen AngleThr AngleDev
OutOfStep_new 10.4207 1375 160.03 180 182.5495
OutOfStep_new 10.4207 1375 180.03 180 181.8143
LDSH_LDFR TimeOut Stage Pshed Qshed ShedLp.u.InitL Voltage Frequency
LDSH_LDFR 11.529 1 13.68 5.97 0.3876 0.97 59.72
LDSH_LDFR 11.633 1 7.92 2.29 0.1646 1.01 59.7
LDSH_LDFR 11.675 1 5.74 2.15 0.6512 0.98 59.71
LDSH_LDFR 11.675 1 2.85 0.91 0.0905 1.03 59.71
LDSH_LDFR 11.687 1 1.99 0.45 0.1585 1.03 59.71
LDSH_LDFR 11.692 1 1.99 0.45 0.1654 0.97 59.71
LDSH_LDFR 11.7 1 1.5 0.35 0.1307 1.03 59.71
LDSH_LDFR 11.721 1 2.13 0.61 0.0392 1.03 59.71
LDSH_LDFR 11.733 1 5.39 1.12 0.2122 1 59.71
VTGTPA TimeOut Pgen Qgen
VTGTPA 11.737 50 30
LDSH_LDFR 11.754 1 3.28 0.79 0.4316 0.99 59.71
LDSH_LDFR 11.762 1 4.58 1.01 0.237 1.01 59.72
LDSH_LDFR 11.767 2 7.1 2.04 0.1304 1.03 59.72
LDSH_LDFR 11.846 1 23.63 0.07 0.6513 1.01 59.65
LDSH_LDFR 11.896 1 14.24 4.68 0.6513 0.99 59.68
LDSH_LDFR 11.95 1 92.48 37.97 0.6513 1.02 59.66
LDSH_LDFR 12.008 1 0.51 0.1 0.0158 0.98 59.66
LDSH_LDFR 12.008 1 0 0 0.3372 1.02 59.65
LDSH_LDFR 12.008 1 19.51 5.69 0.3372 1.02 59.65
LDSH_LDFR 12.008 1 58.65 17.11 0.3372 1.02 59.65
LDSH_LDFR 12.008 1 78.16 22.8 0.3372 1.02 59.65
LDSH_LDFR 12.058 1 16.67 3.34 0.3799 1.02 59.67
LDSH_LDFR 12.062 2 11.91 2.39 0.2714 1.03 59.67
LDSH_LDFR 12.079 1 23.44 2.57 0.4162 1.02 59.67
LDSH_LDFR 12.083 1 1.31 0.18 0.1267 0.97 59.68
LDSH_LDFR 12.083 1 0.91 0.13 0.1267 0.97 59.68
Freq
uen
cy a
t D
iffe
ren
t Lo
ad B
use
s
Steady-State Cascading Outage Analysis
using TransCare
21
Used TransCARE built-in logic for automated breaker placement.
The breaker locations are utilized by TransCare to identify the
Protection and Control Groups (PCGs) that a system component
belongs to.
During contingency solution, the dispatch algorithm restores
generation-load balance in the system following the outage of a
generating unit(s).
Provided generation units participation factors are used to
determine the new redispatch value for each generating unit.
TransCare Flowchart
22
First Initiating
Event
Solve Power
Flow
Disconnect
Loads
Load Bus
Voltages <
Threshold?
Trip
Generators
Generator
Bus Voltages
< Threshold?
Largest
Circuit Overload
> Threshold?
Trip PCG
Containing
Circuit
Last Initiating
Event? Stop
Next Initiating
Event
Start
Yes
No
Yes
No
Yes
No
No Yes
Bus Voltages <
“Voltage
Collapse”
Threshold?
Drop Load @
Affected Buses
During the solution
Yes
Continue solution
No
Shut down island if 10% deficiency
or excess of island generation
Transmission overload tripping:
130% for circuit thermal limits
Low voltage generation units trip:
0.85 pu
Low voltage load trip: 0.85 pu
Summary of Results
23
No. of initiating events: 620
No. of Non-solved cases: 6 (1%)
No. of Capacity Deficiency cases: 0%
No. of Divergent cases: 0%
No. of Non-solved cases after several power flows: 2 (0.3%)
No. of Severe Cases with load loss or cascading events: 388 (62%)
No. of normal cases: 224 (36%)
Critical Event Corridor Identified
PCG1 PCG2 Frequency
PCG04490 PCG04489 5
PCG04026 PCG04490 3
TransCARE Cascading Outage Results
24
0 100 200 300 400 500 600 7000
500
1000
1500
2000
Load loss,
MW
0 100 200 300 400 500 600 7000
1
2
3
Contingency No.
No.
of
line t
rippin
g
Other On-going Efforts
Stakeholder outreach
Parallelized DCAT using distributed processors and
demonstrated on PNNL supercomputing facility
Enhanced the DCAT code to make it user friendly, a manual is
available with several test system examples
Updated DCAT for PSS/E version 33
Improved protection modeling and corrective actions (GMLC
Extreme Events study)
Refine protection settings for distance relays and model Zone 3
protection
Use GAMS for corrective actions optimization
Developing DCAT for Cascading-Outage Analysis for the
Western Interconnection using GE PSLF (BPA TI project)
January 13, 2017 | 25
Acknowledgements
This project is funded by the U.S. Department of Energy Office of Electricity Delivery
and Energy Reliability (DOE-OE). The project team wants to especially thank Mr. Gil
Bindewald, Program Manager and Dr. David Ortiz, Deputy Assistant Secretary for
Energy Infrastructure Modeling and Analysis (OE-40) for their continuing support,
help, and guidance.
The project team appreciates contributions, guidance, and discussion time on a
biweekly basis from José Conto and Sun Wook Kang of the Electric Reliability
Council of Texas (ERCOT).
The project team appreciates technical support in developing the DCAT from
Siemens Power Technologies staff Hugo Raul Bashualdo, Dinemayer Silva,
Jayapalan Senthil, James Feltes, Joseph Smith, and Krishnat Patil. We would like
also to thank Siemens subcontractor Eli Pajuelo for his help in protection modeling
in PSS/E.
The project team appreciates technical support in using TransCARE from Murali
Kumbale from Southern Company and Anish Gaikwad from the Electric Power
Research Institute. Special thanks to the Electric Power Research Institute for
providing a royalty-free copy of TransCARE to be used in this study.
26 January 13, 2017 |
Questions?
January 13, 2017 27
Disclaimer: PNNL power industry partners in this project played only
an advisory role. All models built and simulations were performed by
PNNL staff. Consequently, all questions regarding this presentation
should be only directed to the PNNL PI or PM with the contact
information below. Samaan NA, JE Dagle, YV Makarov, R Diao, MR Vallem, TB Nguyen, LE Miller, B
Vyakaranam, S Wang, FK Tuffner, and MA Pai. 2015. Dynamic Contingency Analysis Tool –
Phase 1 . PNNL-24843, Pacific Northwest National Laboratory, Richland, WA.
http://www.pnnl.gov/main/publications/external/technical_reports/PNNL-24843.pdf
Nader Samaan, PhD, PE (PI) Senior Power Systems Research Engineer Electricity Infrastructure/Analytics Pacific Northwest National Laboratory Richland, WA 509-375-2954 [email protected]
Jeff Dagle, PE (PM) Chief Electrical Engineer and Team Lead Electricity Infrastructure/Transmission System Resilience Pacific Northwest National Laboratory Richland, WA 509-375-3629 [email protected]